DroneTank: Planning UAVs' Flights and Sensors' Data Transmission under Energy Constraints

Runqun Xiong, Feng ShanSensors 2018 · DOI

TL;DR

This paper tackles the complex problem of optimizing data collection for Unmanned Aerial Vehicles (UAVs) acting as mobile data sinks for ground sensors, all while managing strict energy constraints for both the UAV and the sensors. They propose a novel approach, "DroneTank," that decouples the problem into UAV flight planning and sensor transmission scheduling, achieving near-optimal data collection through dynamic programming and a unique "water-tank" method.

What Problem Does This Paper Address?

In many modern applications, like smart grids or precision agriculture, ground-based sensors collect valuable data. However, these sensors might be in remote or difficult-to-access locations. Unmanned Aerial Vehicles (UAVs) offer a promising solution as mobile data collectors, flying to these sensors to gather information. The challenge lies in efficiently coordinating the UAV's flight path and speed with the sensors' data transmission schedules, all while operating within limited energy budgets for both the UAV and the sensors. This joint optimization problem is complex because the UAV's trajectory affects sensor transmission efficiency (due to distance and time), and the sensors' transmission strategies (when and how powerfully to transmit) impact the UAV's available time and energy. Existing approaches often struggle with high computational complexity or make simplifying assumptions about sensor deployment (e.g., linear arrangements).

Methods

The authors propose a novel framework called "DroneTank" that breaks down the joint UAV flight planning and sensor transmission scheduling problem into two manageable subproblems:

  1. UAV Flight Planning (UAV-FP): This subproblem focuses on determining the optimal flight path and speed for the UAV. They simplify this by reducing it to finding the optimal number and locations of "turning points" (where the UAV might slow down or change direction). A dynamic programming approach is used, guided by three key observations:
  2. Fly Closer: Sensors transmit more efficiently when the UAV is closer.
  3. Fly Slower: A slower UAV provides more time for sensors to transmit data.
  4. More Energy, More Important: Sensors with larger energy budgets are prioritized. The UAV-FP problem is formulated to minimize a combined "error" metric (representing deviations from these observations) while respecting the UAV's energy budget.

  5. Sensor Transmission Scheduling (SEN-TS): Once the UAV's flight plan is determined, this subproblem focuses on optimizing when and how powerfully each sensor transmits. A novel technique called the "water-tank" method is introduced. This method, inspired by water-filling algorithms but adapted to handle power constraints, models the problem of allocating transmission power to sensors within their available time slots and energy budgets to maximize data throughput. This is then integrated with dynamic programming to find the optimal slot assignments and power controls for all sensors.

Datasets

Not discussed in the paper. The evaluation uses simulated data based on defined parameters.

Key Insights and Ideas

  • Decoupling is Key: The paper demonstrates that by carefully observing the interplay between flight planning and transmission scheduling, the complex joint problem can be effectively decoupled into two more tractable subproblems, leading to a near-optimal solution.
  • The "Water-Tank" Analogy: The water-tank method provides an elegant and intuitive way to solve the sensor transmission scheduling problem. It visualizes filling "tanks" (representing transmission power in time slots) with a limited "water" supply (sensor energy), ensuring optimal power allocation under constraints.
  • Prioritizing Energy: The observation that sensors with larger energy budgets are more important for data collection is crucial for guiding the UAV's flight planning, ensuring that valuable data from these sensors is prioritized.
  • Trade-offs in Flight Planning: The paper highlights the inherent trade-offs in UAV flight planning: flying closer and slower improves sensor transmission but consumes more UAV energy. The proposed method balances these by minimizing an "error" metric that reflects these trade-offs.

Key Results

The proposed "DroneTank" algorithm, which combines dynamic programming for UAV flight planning and the water-tank method for sensor transmission scheduling, achieves near-optimal data collection performance. Simulations show that the algorithm consistently collects approximately 95% of the maximum possible data, even when varying key parameters like network size, sensor energy budgets, UAV energy budgets, UAV speed, transmission range, and turn energy costs. The performance is shown to be stable across different network sizes and sensor energy budgets. The UAV energy budget and speed were found to have a greater impact on performance than other parameters.

Limitations and Future Work

  • Line-of-Sight Channel Assumption: The current model assumes a line-of-sight wireless channel. Future work aims to incorporate more sophisticated channel models, such as air-to-ground channels, to better reflect real-world scenarios.
  • Simplified UAV Propulsion Model: The UAV propulsion energy consumption model is simplified. Future research will focus on developing more accurate models based on multi-rotor UAV characteristics and real-world measurements.
  • 2D Flight Planning: The current work assumes 2D flight planning at a fixed altitude. Future work will explore optimal altitude selection for enhanced coverage and energy-efficient flight planning.
  • Multi-UAV Systems: The paper focuses on a single UAV. Extending the framework to multi-UAV systems for broader applications is a promising future direction.

Summary updated

BibTeX
@article{xiong2018dronetank,
  author = {Runqun Xiong and Feng Shan},
  title = {{DroneTank: Planning UAVs' Flights and Sensors' Data Transmission under Energy Constraints}},
  journal = {Sensors},
  year = {2018},
  doi = {10.3390/s18092913},
}
Figure